X-Message-Number: 15319
Date: Thu, 11 Jan 2001 07:50:00 -0500
From: Thomas Donaldson <>
Subject: superficial ideas about computers

Again, for Joseph Kehoe:

Unfortunately there are problems with your ideas. Perhaps they can be
solved, perhaps not... but here goes.

First, it's not at all clear that increasing the number of neurons 
could simply involve growth of new ones from old. The essential problem
is that such new neurons must connect to some other neurons, which in
turn connect with others, etc. They cannot simply grow, they must
connect ... and if they don't they will die. Furthermore, neurons 
also connect over long distances, so that the claim that they only
connect with nearby neurons is false (though to be fair, this happens
much less often than connections with nearby neurons, at least if we
consider those newborn in adult life). 

As for a model of neurons which allows them to increase, yes, there
are such models. The real issue is that of just how they increase; it's
worth pointing out here that multiplication of neurons in real brains
was accepted much more recently than the one model you mention. That
model shows little similarity to the real workings of brains, and 
thus I put little weight on it.

You also suggest that computer systems could be designed to be capable
of self repair. The self-repair you suggest remains highly restricted
in terms of what it can do. We get our own materials and process them;
and to some extent (admittedly right now less in brains, but I wasn't
talking only of brains) plan out just how we do that repair. We produce
new neurons and their connections from materials which are not neurons.
In general, our self-repair goes much farther than any machine yet
built.

Finally, for Turing machines, you seem to ignore the major factor of 
TIME. Even if everything I have said so far in this message could be
rejected, the problem of the importance of TIME will remain. Each of 
the millions of neurons in our brain is itself a small computer, with
a small memory. The reason for this is simple: a single computer does
not exist that can carry out the required processing fast enough to
do the job of each of these neurons ... however small and weak they
may seem to be. Turing did not but should have considered TIME as
a variable in his work ... but then the need for computers fast
enough to solve problems in 30 minutes which would have taken 
millions of years for a single processor simply hadn't yet arisen.

It is one thing to proposdevice capable of doing what we human
beings do with our brains and feelings, and quite another to build 
one. That is one of the hardest issues this question raises. Sure,
we can come up with very rough notions about computers which (in
our thoughts alone) would allow them to do what human beings do.
But when we examine them they become more and more complex, and 
more and more difficult, and many questions arise. The basis for
believing that computers for a model for human beings comes from
unexamined assumptions which deserve close examination, none of which
they have received.

		Best and long long life for all,

			Thomas Donaldson

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=15319